Explore AI
Everyday writing and chat
Summarising, drafting, explaining, thinking out loud. The least demanding use.
Summarising, drafting, explaining, thinking out loud. The least demanding use.
What we would try first
- Qwen3 8B 8 GB at Q4 The default answer for most people with 16 GB or a 12 GB card. Start here, and only look elsewhere if it fails at something specific.
- GPT-OSS 20B 16 GB at Q4 A Mixture-of-Experts model: 20B of weights but only about 3.6B active per token. That is why it fits in 16 GB and still runs quickly. The strongest argument for MoE on consumer hardware.
- Llama 3.1 8B Instruct 8 GB at Q4 Enormously widely supported, which matters more than benchmark scores when you are troubleshooting. The licence restricts very large-scale commercial use.